Property Valuation API - AVM, Rent Estimate, Sold Prices

Value any European residential address for sale or for rent with a confidence range, and pull the listings, recorded sold prices, market statistics and neighbourhood scores around it from the same actor.

What this actor does

• Automated valuation for sale or rent. Paste addresses, get a sale price with a lower and upper bound, or a gross and a net monthly rent each with their own bounds. Every valuation carries the provider's confidence grade, the coordinates it resolved the address to, and a location score.
• Value the same property at several dates. Give a list of valuation dates and get one row per property per date, which is how to measure what a portfolio has done rather than only what it is worth today.
• Live listings around any point. Asking price and price per square metre, days on market, listing start and end dates, price change count, agency contact, images, and a flag saying whether the provider placed the listing at its real address or only in the right area.
• Recorded sold prices. Actual registered transactions with the date, price, price per square metre and per square foot, plus the property attributes on file including land area, bedrooms, bathrooms, leasehold status, energy label and planning use.
• Market statistics and percentile ranks. Mean, count and any percentile of price, rent, area, rooms or days on market, grouped by month, quarter, year or trailing 90 days. Ask for the percentile rank of a specific price and the provider tells you where it falls in its market.
• Location scores. Nine measures for a point - overall location, catering, family, health, leisure, shopping, view, noise and nuisance - each between 0 and 1 and each oriented so that higher is better.
• Demographics by administrative division. One socio-economic series at a time, returned per division, year and subtype with both the absolute figure and its share of the division.
• Pay per result. You are charged for rows that carry a real record. Addresses the provider cannot place are free.

Bring your own credentials

This provider authenticates with an account username and password rather than an API key, and exchanges them for an access token that lasts twelve hours. Paste yours into the API username and API password fields, or set them as the DATA_API_USERNAME and DATA_API_PASSWORD secrets. If you keep one combined secret across actors, DATA_API_KEY set to username:password also works.

Both halves are stored as secrets. The actor signs in once per run rather than once per request, because the provider rate limits repeated authentication specifically, and it refreshes the token silently if a long run outlasts it.

Coverage is licensed per country

Every request names the country it is about, and your account is licensed for the countries you bought. A country outside your plan comes back as a refusal rather than as an empty result, and the actor reports that as a plan problem rather than a bad credential.

Example output

{
  "recordType": "valuation",
  "requested": "Gerechtigkeitsgasse 19, 8001 Zurich",
  "found": true,
  "valuationDate": "2026-09-01",
  "dealType": "sale",
  "salePrice": 1845000,
  "salePriceLower": 1705000,
  "salePriceUpper": 1985000,
  "currency": "CHF",
  "confidence": "good",
  "locationScore": 0.91,
  "latitude": 47.36925,
  "longitude": 8.531118,
  "statusCode": 200,
  "statusMessage": "OK"
}

FAQ

How accurate is an automated valuation?

Every valuation comes with a lower and upper bound and a confidence grade of poor, medium or good, so you can judge it rather than trust it. A good grade with a tight range means the provider had comparable evidence nearby; a poor grade with a wide range means it did not. Supplying the building year, living area and room count materially narrows the range, because those are the strongest inputs the model has.

What is the difference between gross rent and net rent?

Gross rent includes the service charges a tenant pays on top of the base rent; net rent excludes them. They are separate fields in the output because they are different figures, and which one a local market quotes varies by country. A rent valuation returns both, each with its own range.

Why does the output not have a single "value" column?

Because the provider's own value field is deprecated and shadows three different numbers - the sale price, the gross rent and the net rent. Reading it would work today and silently return the wrong figure once it is withdrawn, so this actor reads each price by its own name and keeps them in separate columns.

How do I get sold prices rather than asking prices?

Use recorded sale search. Listings are what a property was advertised at; recorded sales are what it actually changed hands for, taken from the registry. The two answer different questions, and comparing them for the same area is how you measure the gap between asking and achieved. Recorded sale coverage is strongest in Austria, France and the UK.

Can I tell whether an asking price is high for its area?

Yes, in one request. Use listing statistics, choose the percentile rank type, and give the asking price as the percentile value. The provider answers with where that price falls in the distribution, so 78 means it is higher than 78 percent of comparable listings.

How do I build a price time series for a neighbourhood?

Use either statistics mode, group by the transaction date or the listing start date, and set the grouping period. Trailing 90 days smooths a thin market better than a calendar month does, because a month with four sales in it produces a median that moves for no real reason.

Are the listing coordinates exact?

The row tells you. The preciselyLocated flag is true when the provider placed the listing at its real address and false when it could only place it in the right area. Mapping an approximate pin as though it were exact is the usual mistake with listing data.

Why are the noise and nuisance scores high in quiet places?

All nine location scores are oriented the same way, so a higher number is better on every one of them. A noise score of 0.95 means the location is quiet, not that it is loud. They are also read from the provider's nested score block rather than from the top level copy, so a score somebody has overridden on their own account is reported as overridden instead of passing as a measured one.

How many properties can I value in one request?

The provider caps a call at fifty valuations, counted as properties multiplied by dates. So fifty addresses on one date is one request, and the same fifty on three dates is three. The actor works that out and batches accordingly; you set the addresses and the dates and it sizes the requests.

What does it cost?

Pricing is pay per result: 0.018 USD per valuation, 0.01 USD per listing or recorded sale, 0.01 USD per location scored, and 0.008 USD per market statistic or demographic figure, with Apify platform usage included. Rows the provider found nothing for are not charged. You also need your own account with the data provider, which bills on its own plans.

Does the actor fail if my credentials are wrong?

No. Missing credentials, a rejected username or password, an endpoint or country outside your plan, or an empty input all end the run cleanly with a message saying what to change. Only a real outage or a bug fails the run.

Keyword map

property valuation API, AVM API, automated valuation model, European property data API, rent estimate API, sold house prices API, property price per square metre, real estate market statistics API, Swiss property data, German property data, French property data, UK sold prices, property listings API Europe, days on market, location score API, neighbourhood scores, socio-economic data API, property confidence range, gross rent net rent